Maximal polynomial modulations of singular Radon transforms (Q6187693)

In the paper, \(L^2-L^p\) estimates, \(1\le p<2\), on the torus \(\mathbb T^d\simeq [-1/2,1/2)^d\), \(d\ge1\), for maximal polynomial modulations of Calderón-Zygmund (C-Z for short) operators in the setting of anisotropic scaling are proved. Given a vector of exponents, \(\alpha=(\alpha_1,\ldots,\alpha_d)\in \mathbb N^d\), \(\alpha_1\le\ldots\le\alpha_d\), the anisotropic dilations \((\delta_r)_{r>0}\) with exponent \(\alpha\) are \(\delta_r(x)=(r^{\alpha_1}x_1,\ldots,r^{\alpha_d}x_d)\), and the anisotropic distance function \(\rho\) is \(\rho(x)=\inf\{r>0\colon |\delta_{1/r}(x)|\le1\}\). A C-Z kernel in this setting is a \(K\in\mathcal{S}'(\mathbb R^d)\) that outside 0 coincides with a function satisfying natural C-Z conditions, where the Euclidean norm \(|\cdot|\) is replaced by \(\rho\) and the length of \(\alpha\), \(|\alpha|:=\sum_1^d \alpha_i\), replaces the dimension \(d\). Then \(Tf:=f\ast K\), \(f\in\mathcal{S}(\mathbb R^d)\), defines the C-Z operator associated with \(K\), bounded on \(L^2(\mathbb R^d)\). An important ingredient of the considered framework is the assumption on an admissible decomposition \(K=\sum_{s\in\mathbb Z}K_s\) of the kernel \(K\), that includes the \(L^2(\mathbb T^d)\to L^2(\mathbb T^d)\) boundedness of two appropriately associated operators, the maximally truncated singular integral \(R^K\) and the maximal average operator \(M^K\). Let \(\mathcal{Q}\) stand for the linear space of all real polynomials on \(\mathbb R^d\) of degree \(\le d\) and without constant term. The maximal polynomial modulation operator \(T^\mathcal{Q}\) is defined for \(f\in L^2(\mathbb T^d)\) by \[ T^\mathcal{Q}f(x)=\sup_{Q\in \mathcal{Q}}\sup_{0<\underline{R}<\overline{R}}\Big|\int_{A(\underline{R},\overline{R})}K(x-y)e^{iQ(y)}f(y) dy\Big|, \] where \(A(\underline{R},\overline{R})=\{y\in x+\mathbb T^d\colon \underline{R}<\rho(x-y)<\overline{R} \}\). The main result of the paper extends the main result of \textit{P. Zorin-Kranich} [Adv. Math. 386, Article ID 107832, 40 p. (2021; Zbl 1467.42024)] to the above-described setting and says that \(T^\mathcal{Q}\) is bounded from \(L^2(\mathbb T^d)\) to \(L^p(\mathbb T^d)\), \(1\le p<2\); moreover, the constant involved in this bound depends, for \(d\), \(p\), \(\alpha\) fixed, on \(\|R^K\|_{2\to2}\), \(\|M^K\|_{2\to2}\), and the constant entering the relevant bounds of the C-Z kernel \(K\).